The same drawing means different things to different people
Where ambiguity survives even a fully dimensioned drawing.
A drawing with every dimension filled in and every tolerance stated can still mean different things to different people, because a dimension only answers "how big," while a real part also depends on many smaller, unstated decisions, which surface counts as the reference, how sharp a corner may be, what condition an edge should be left in, that adding more numbers to the page does not automatically resolve.
Shorthand needs shared convention
The previous articles in this set treated the numbers as the whole subject, worst-case against statistical stacking and the cost of a tighter band, and both quietly assumed the drawing was otherwise unambiguous. That assumption is often wrong. A drawing is a compressed, shorthand description of a part, and shorthand relies on shared convention to fill in everything it leaves unsaid. Two readers who bring slightly different conventions to the same page can walk away with different parts in mind from the identical set of numbers.
A recipe that leaves out the technique
A recipe listing every ingredient to the gram can still produce very different dishes in two kitchens, because it rarely says how finely to chop the onion, how hot "medium heat" runs on a particular stove, or how long "until golden" takes. Nobody calls the recipe wrong for leaving those out, since a recipe spelling out every technical decision would be unreadable, yet the gap between what is written and what is needed to reproduce the dish is real, and assumption fills it.
A professional kitchen closes that gap by training every cook in the same house techniques, the same knife cuts, the same reading of "medium heat" on its own stoves, so the recipe's gaps get filled the same way whoever is cooking that night. An engineering drawing leaves an equivalent gap, and it is filled correctly only while every reader shares the conventions the drawing was written against.
Where the gaps hide on a real drawing
An unflagged surface carries no explicit finish requirement, which one reader takes to mean whatever finish the process naturally leaves and another takes to mean smooth enough to look deliberate. A corner with no radius or chamfer callout is assumed sharp by one shop and broken slightly by another, each following its own standing habit. A fit between a shaft and a bore described only as "sliding," without a clearance class, a thread engagement length left unstated, or two functionally identical faces with no indication of which one is cosmetic all carry the same hidden gap.
The datum scheme and measurement method from earlier in this set live here too. A drawing can specify a datum surface and a tolerance perfectly and still leave open which instrument, technique and conditions are meant to verify it, the very ambiguity that turned two honest measurements into an argument two articles ago. A drawing with no blank fields has passed a much lower bar than a drawing with no open questions, and a review that hunts for unstated assumptions instead of checking that every expected number is present usually turns some up.
What a standard settles, and what it cannot
Well-run engineering organisations do not leave most of this to chance. A standard for geometric dimensioning and tolerancing exists to give the unstated defaults a formal, agreed meaning, so that nothing depends on whichever convention a reader carried in from a previous job. A drawing produced under a clearly stated standard, with default tolerances, default surface treatment and a defined datum reference frame set once in the title block, closes most of the gaps described above before anyone has to guess.
Some ambiguity survives, since a standard only covers the categories its authors thought to anticipate. Two organisations each applying a standard rigorously can still read a drawing differently if they follow different standards, a supplier trained under one national convention interpreting a default tolerance slightly differently from a customer trained under another. That is why a drawing crossing between organisations often names the specific standard and edition it was drawn to.
On the drawing itself, the remaining gaps are closed with explicit notes for the categories that tend to hide them: a default surface finish and edge-break callout applying to every feature unless overridden, and a stated datum and inspection method. During review the useful question goes beyond whether every dimension is present, to what two reasonable people reading the drawing independently would disagree about. That question is harder to answer honestly, and it catches what a completeness checklist walks straight past. A drawing is finished when a stranger reading it cold would build the same part the designer had in mind.